Velvet — AI-Powered Valet Management App
Tech Stack
Overview
Velvet is a two-sided mobile platform connecting valet businesses (restaurants, hotels) with their valet attendants. When a customer hands over their car, the valet takes a photo — the backend AI pipeline automatically identifies the vehicle's license plate, brand, and color in seconds. The business owner monitors everything in real time through a dashboard with statistics, payment tracking, and device management.
The app features a Business Panel with real-time dashboard, ticket management, AI-assisted check-in, smart checkout with nearest valet assignment via FCM, payment tracking, parking lot management with GPS, and analytics. The Valet Panel includes open tickets view, AI camera for auto-fill, map navigation, QR code scanning, and push notifications.
Challenges
- Building a two-sided platform with distinct workflows for business owners and valet attendants
- Integrating AI-powered vehicle identification (plate, brand, color) with the camera flow
- Implementing real-time nearest-valet assignment using geolocation and FCM push notifications
Solutions
- Scope-based JWT authentication with separate business and valet panels and GetX routing
- Connected Flutter camera with FastAPI backend AI pipeline for seamless vehicle identification
- Used Haversine formula on backend for proximity calculation and FCM for instant valet alerts